Classifying vertical facial deformity using supervised and unsupervised learning.

OBJECTIVES: To evaluate the potential for machine learning techniques to identify objective criteria for classifying vertical facial deformity. METHODS: 19 parameters were determined from 131 lateral skull radiographs. Classifications were induced from raw data with simple visualisation, C5.0 and Kohonen feature maps; and using a Point Distribution Model (PDM) of shape templates comprising points taken from digitised radiographs. RESULTS: The induced decision trees enable a direct comparison of clinicians' idiosyncrasies in classification. Unsupervised algorithms induce models that are potentially more objective, but their blackbox nature makes them unsuitable for clinical application. The PDM methodology gives dramatic visualisations of two modes separating horizontal and vertical facial growth. Kohonen feature maps favour one clinician and PDM the other. Clinical response suggests that while Clinician 1 places greater weight on 5 of 6 parameters, Clinician 2 relies on more parameters that capture facial shape. CONCLUSIONS: While machine learning and statistical analyses classify subjects for vertical facial height, they have limited application in their present form. The supervised learning algorithm C5.0 is effective for generating rules for individual clinicians but its inherent bias invalidates its use for objective classification of facial form for research purposes. On the other hand, promising results from unsupervised strategies (especially the PDM) suggest a potential use for objective classification and further identification and analysis of ambiguous cases. At present, such methodologies may be unsuitable for clinical application because of the invisibility of their underlying processes. Further study is required with additional patient data and a wider group of clinicians.

Identification of matrix metalloproteinases and their tissue inhibitors type 1 and 2 in human masseter muscle.

Changes in masticatory muscle structure and function are either developmental, as seen in anomalies of facial form, or adaptive, as seen during procedures such as orthognathic surgery and functional-appliance orthodontic therapy. Remodelling of muscle extracellular matrix is pivotal in these processes. This turnover is mediated via members of the family of enzymes known as matrix metalloproteinases (MMP) and inhibited by the tissue inhibitors of metalloproteinases (TIMP). The aim here was to investigate the in vivo pattern of expression and distribution of MMPs and TIMPs in masseter muscle of humans with both normal and abnormal facial forms. Masseter muscle biopsies were taken from 10 patients, four with long-face syndrome and six normal controls as confirmed by cephalometry. Immunohistochemical techniques were used to show the pattern and distribution of MMPs and TIMP proteins in the muscle. Zymography of tissue extracts was used to determine the presence of MMP activity. Reverse transcriptase-polymerase chain reaction (RT-PCR) was used to detect the presence of MMP and TIMP-2 mRNA. MMP-1 was expressed around the individual muscle fibres, especially in those fibre surfaces in contact with the interstices of the connective tissue and around blood vessels. MMP-9 staining was less intense and was expressed in the interstices of the connective tissue and around blood vessels. Zymography of protein extracts confirmed that MMP-9 activity was present. MMP-2 and MMP-3 were not expressed in the samples, although MMP-2 mRNA could be detected by RT-PCR and its activity could be detected by zymography. Intense TIMP-1 staining was present around each muscle fibre, in the interstices of the connective tissue and surrounding blood vessels; TIMP-2 mRNA could be detected in all samples. These staining patterns were seen in all biopsies examined and were irrespective of the facial form of the donor. These findings provide evidence that the mechanisms required for matrix remodelling are present in the human masseter muscle.